Local learning‐based model‐free adaptive predictive control for adjustment of oxygen concentration in syngas manufacturing industry

Abstract

During the united gas improvement (UGI) gasification process in the syngas industry, the oxygen-enriched technique plays an important role, since the obtained oxygen-enriched air with a high oxygen concentration can enhance the production efficiency of the syngas. However, satisfactory control performance for the oxygen concentration of the oxygen-enriched air is hard to achieve because an accurate dynamical model of the oxygen concentration control process by the first principles is fairly difficult to obtain due to strong non-linearity and unknown disturbances in practice. A novel data-driven control method called compact-form-dynamic-linearisation-based model-free adaptive predictive control approach combined with the local learning (LL-CFDL-MFAPC) is proposed to address the control problem. In LL-CFDL-MFAPC, the online and offline input-output measurement data of the plant are fully and simultaneously utilised during the control process, and the design of the controller is model free by means of compact-form-dynamic-linearisation technique. Moreover, the controller has strong robustness because the prediction mechanism participates in control design and only the input/output measurement data are used. The stability and convergence of LL-CFDL-MFAPC are guaranteed by theoretical analysis under several reasonable assumptions, and simulation experiments using real data collected from a practical UGI gasifier verify that the oxygen concentration control problem can be effectively addressed by the proposed method.